In architecture design, while Generative AI can effortlessly create initial prototypes, architects struggle to update designs to stakeholders’ evolving requirements. Through a formative study (N = 12), we identified specific obstacles that architecture designers face when meeting the dynamic design demands of various project stakeholders. We therefore developed ArchiConnect, a proof-of-concept interactive system that helps architects communicate with multiple stakeholders and update final design deliverables. ArchiConnect supports creativity and engagement by visualizing evolving demands, conflicts, and concept extractions from diverse stakeholders. We evaluated our system in a week-long user study (N = 8) with a simulated project. Participants found ArchiConnect effective for improving multi-stakeholder communication and management, describing it as intuitive and useful. Our findings offer design considerations for future AI tools to better handle dynamic stakeholder needs, including how to address sustainability requirements in line with development goals.

• The study emphasizes knowledge graph-powered product iteration within an under-explored NPD domain of newer and less-established novel products. • An interpretable conceptual design knowledge graph (I-CDKG) is constructed to facilitate designers in generating innovative and cost-effective heuristic product ideations. • A hybrid method combining deep-learning ERNIE-BiGRU-CRF model, BIESO labeling mode, and triple-extracting algorithm is proposed to facilitate the I-CDKG construction. • The I-CDKG boasts both inherent and acquired interpretability reinforced by a Cluster-Relation-Nest organizational strategy for the intuitive locating of design knowledge. Novel products emerge over time to survive the competitive landscape as no existing product can perpetually satisfy all evolving customer expectations. These products are often characterized by groundbreaking solutions previously unavailable on the market. However, the swift imitation of successful novel products by competitors underscores the need for sustained iteration and continuous improvement. Designers increasingly face challenges in keeping up to date with the growing volume and fragmented nature of design information from diverse sources. While knowledge graphs show promise in structuring and organizing complex design information, their effective application in the ideation process remains limited due to difficulties in automatic knowledge extraction and the lack of interpretability aligned well with designers’ cognitive processes. This study proposes an integrated method to construct an interpretable conceptual design knowledge graph (I-CDKG) that features both inherent and acquired interpretability for heuristic product ideation. First, the schema layer models product design knowledge and governs the semantic connection of design information reinforced by design cognition principles to create a reasonable organizational framework to foster intuitive knowledge exploration. Second, the data layer mainly fulfills automatic and smooth design knowledge extraction for I-CDKG construction through the deep learning ERNIE-BiGRU-CRF model combined with BIESO labeling mode and triple-extracting algorithm. Third, the application layer empowers designers to visually delve into interpretable design knowledge to locate inspiration from cluster, relation, and nest levels and enable constant I-CDKG expansion as design schemes proliferate. A case study on the smart cat litter box demonstrates the feasibility of the proposed methodology. The evaluation results confirm the I-CDKG’s advantages as a productive design tool for inspiring creative, practical, and cost-effective product ideations, thereby empowering the iterative development of competitive novel products.

The rapid increase in the number of individuals with Autism Spectrum Disorder (ASD) has drawn extensive attention from both the general public and researchers. Artificial Intelligence (AI) has been applied in the assessment, early diagnosis, and intervention of ASD to enhance the efficiency of clinicians and reduce tension in medical resources. However, the adoption of AI systems in clinical practice is relatively limited due to the challenge of complexity and diversity of ASD. Thus, involving insights into clinicians' perceptions and barriers toward the role of AI is crucial for enhancing clinicians-AI cooperation for autism. Through conducting the semi-structured interview with 18 physicians across tertiary and secondary hospitals in various regions, this study indicates the positive attitude toward collaborating with AI among physicians. Additionally, some concerns are also reported, such as the complexity of ASD, uncertainty of AI capabilities, and understandability of AI. The findings of this study highlight the significance of human-centered AI in satisfying different stakeholders' needs and discuss the potential implications of AI capabilities for adopting AI in future autism research.